Research of Equipment Fault Diagnosis Based on PHM High Performance Computing Platform

Aiming at the problems of poor real-time fault diagnosis and low efficiency in the complex equipment PHM engineering maturity, a fault diagnosis implementation scheme based on PHM high performance computing platform is proposed. The BP neural network algorithm is used as an example to verify. Firstly, the current technical status and urgent needs of the existing PHM operation platform are analyzed. The overall structure and software and hardware optimization configuration of PHM high performance computing platform with FPGA and DSP as the core are expounded. Then, by means of module division, HDL design, functional verification and package testing of the time domain feature extraction method and BP neural network, the implementation of the platform fault diagnosis algorithm is carried out. Finally, combined with the analysis of a certain type of on-board fuel pump fault data, comparative analysis was carried out with the CPU platform operation. The results show that the fault diagnosis implementation proposed in this paper has high real-time performance, low resource consumption and low power consumption, which can provide an important reference for complex equipment PHM engineering applications.